Cancer clinical trials in the USA: patient eligibility, generalizability of results and technology transfer.

The impact of changes in the source population of patients for cancer clinical trials in the USA is considered. The first change, which has already taken place, is the recent trend of including patients treated of community hospitals in cooperative group and other randomized trials. This change was designed to augment accrual to studies, and to disseminate state-of-the-art information on patient care. The change has been shown to be a successful one, with community hospitals providing high quality participation in the trials. In addition, the research protocols have been seen to influence the care of many patients treated off-protocol. The inclusion of community hospitals in the trials clarifies our understanding of the generalizability of the results of the trials. The issue of generalizability is, however, obscured by the fact that the great majority of cancer patients are not treated on protocols. Indeed most of them are ineligible for protocols. The rationale for strict eligibility is unclear in many cases, and its use greatly reduces accrual and casts doubt on the generalizability of results. In the future, more care should be paid to the selection of the source population for clinical trials.